Secure framework for IoT applications using Deep Learning in fog Computing

Abstract

An efficient offloading strategy provides a promising solution to latency-sensitive applications in a cloud fog environment, but the fog computing environment is not fully protected and fascinates the attackers due to its dynamic and hazardous environments. Several security-based frameworks have been proposed by the authors in an integrated cloud fog environment for IoT applications to mitigate, prevent and defend against vulnerabilities, but failed to address the mentioned issues. We have proposed a lightweight secure framework using Deep Learning (DL) techniques to detect security attacks and monitor the network traffic for IoT applications in this article. The proposed Artificial Neural Network (ANN) based model is trained and deployed over a cloud platform while the detection mechanism is implemented on the fog Nodes to reduce the possibilities of any vulnerabilities and delayed execution. The proposed lightweight framework performance is analyzed and tested over NSL-KDD datasets and compared with the respective baseline techniques Support Vector Machine (SVM), Logistic Regression and Decision Trees to validate the results. We observed from the experimental results that the proposed framework provides the guarantee to defend against vulnerabilities and surpass the baseline algorithms in terms of QoS parameters. Our framework achieved 99.43% accuracy, 99.26% precision and least (0.7396%) False Alarm rate.